Vehicle CAN-Bus Timing Analysis for Masquerade Attack Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current intrusion-detection systems in vehicle CAN-bus networks cannot effectively identify the source of malicious messages, particularly in cases of masquerade attacks, due to the absence of a MAC Address and the complexity of embedded systems.

Innovation Solution

A monitoring method that employs statistical analysis of periodic message arrival times to differentiate between nodes based on confidence intervals and voting mechanisms, allowing for the identification of the electronic control unit transmitting a message, especially during attacks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If statistical analysis of message arrival times is implemented to identify message sources, then the ability to detect malicious messages improves, but the device complexity increases

Engineering Contradiction:
Improvesource identification accuracyVSAvoidmonitoring system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary learning during a first time interval to establish statistical parameters and confidence windows for normal message arrival patterns before actual security monitoring begins. This pre-characterization of legitimate traffic patterns enables subsequent anomaly detection without requiring complex real-time analysis of message sources.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces a monitoring device that acts as an intermediary between the CAN-bus nodes and the vehicle's control systems. This device captures and analyzes message traffic, using statistical methods to identify anomalies while leaving the original nodes unchanged and maintaining their existing communication protocols.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If time-based statistical analysis is used to differentiate nodes, then the ability to detect masquerade attacks improves, but the loss of time for message processing increases

Engineering Contradiction:
Improveattack detection reliabilityVSAvoidmessage processing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system applies statistical analysis selectively rather than to every single message. During the learning phase, it collects sufficient data to establish confidence windows, then during monitoring, it only performs anomaly checks when messages fall outside expected parameters, reducing overall processing overhead while maintaining detection reliability.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent transforms the detection approach from analyzing message content and sources to analyzing temporal parameters - specifically the arrival times of messages. By changing the detection parameter from spatial (which node sent it) to temporal (when did it arrive), the system achieves reliable attack detection with minimal processing overhead.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If confidence interval voting mechanism is implemented to identify attacker, then the precision of attacker identification improves, but the device complexity increases

Engineering Contradiction:
Improveattacker identification precisionVSAvoidanalysis mechanism complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the identification process into distinct phases: a learning phase where statistical parameters are established for each message identifier, and a monitoring phase where anomalies are detected by comparing against pre-established confidence windows. This segmentation simplifies the complexity by preparing reference data in advance rather than performing complex analysis in real-time.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The monitoring device creates copies of message identifiers and their associated statistical parameters during the learning phase, storing them as reference data. During monitoring, it compares incoming messages against these pre-created copies rather than performing complex source analysis, thereby achieving precise attacker identification with reduced computational complexity.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP4224790A1Method for protection from cyber attacks to a vehicle based upon time analysis, and corresponding device
Publication Date: 2023.08.09 MARELLI EURO SPA
  • EP4224790A1 patent drawingFigure 1
  • EP4224790A1 patent drawingFigure 2
  • EP4224790A1 patent drawingFigure 3

AI summary

Described herein is a method for protection from cyber attacks in a vehicle communication network (10), in particular a CAN (Controller Area Network), said network comprising: a communication bus (10), in particular a CAN-bus; a plurality of nodes (11) associated to said communication bus (10) in a signal-exchange relationship and associated at least in part to control units for controlling functions of the vehicle, said nodes (11) exchanging messages (MSG, MSG') passing between nodes of said plurality of nodes (11), said messages (MSG) being distinguished by respective message identifiers (ID), said method (100) comprising, at a control node (20) associated to said communication bus (10): executing a protection and monitoring procedure (100) comprising: a learning step (200), which is carried out during a first time interval and comprises: acquiring (210, 220, 230) sets (vIDk) of values of times of arrival (TSIDk) at the device (20) of messages (MSG) exchanged on the network (10), ordered according to a respective message identifier (IDk), computing (240) one or more statistical parameters ( µIDk, μIDk,sIDk2) of said sets of arrival-time values (TSIDk), and subsequently obtaining (250) statistical parameters (ωIDk) of confidence windows for said one or more statistical parameters ( µIDk, μIDk,sIDk2) of said sets of arrival-time values (TSIDk); and a subsequent classification and anomalydetection step (300), which is carried out during a second time interval and comprises: acquiring (310, 320, 330) sets (uIDk) of values of times of arrival (TS'IDk) at the device (20) of messages (MSG'), computing (340) corresponding one or more statistical parameters ( μ′IDk,s′IDk2) of said sets of arrival-time values (TS'IDk) to obtain votes (V, W) of membership of said one or more statistical parameters ( μ′IDk,s′IDk2) of said sets of arrival-time values (TS'IDk) in confidence windows of the sets of arrival-time values (TS'IDk) acquired during the learning step (300) on the basis of said statistical parameters (ωIDk) of confidence windows for said one or more statistical parameters ( μ′IDk,s′IDk2) of said sets of arrival-time values (TSIDk), and evaluating (350, 360) issuing of malicious-message alarms (AL) on the basis of said membership votes (V, W).